Efficient Indoor Depth Completion Network Using Mask-adaptive Gated Convolution
Authors: Tingxuan Huang, Jiacheng Miao, Shizhuo Deng, Tong Jia, Dongyue Chen
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on popular benchmarks, including NYU-Depth V2, DIML, and SUN RGB-D, demonstrate that our model outperforms stateof-the-art methods in both accuracy and efficiency. Experimental results demonstrate that our model outperforms the state-of-the-art on three popular benchmarks, including NYU-Depth V2, DIML, and SUN RGB-D datasets. We conducted comprehensive experiments on three popular benchmark datasets: NYU-Depth V2, DIML, and SUN RGB-D to validate the performance of the model. |
| Researcher Affiliation | Academia | 1College of Information Science and Engineering, Northeastern University, China 2Foshan Graduate School of Innovation, Northeastern University, China 3National Frontiers Science Center for Industrial Intelligence and Systems Optimization, Northeastern University, China EMAIL, EMAIL, EMAIL |
| Pseudocode | No | The paper describes the methodology using text and architectural diagrams (Figures 2, 3, 4), but no explicit pseudocode or algorithm blocks are provided. |
| Open Source Code | Yes | Code https://github.com/htx0601/Maga Conv. |
| Open Datasets | Yes | NYU-Depth V2 (Silberman et al. 2012) is the most authoritative and widely used benchmark dataset for depth image completion... DIML (Cho et al. 2019) This dataset includes images with typical edge shadows and irregular holes... SUN RGB-D (Song, Lichtenberg, and Xiao 2015) is an extensive dataset comprising 10,335 densely captured RGBD images... |
| Dataset Splits | Yes | NYU-Depth V2... 1449 officially labeled images for evaluation. DIML... We utilize 2000 pairs of labeled samples from the indoor part of the datasets according to the official split. SUN RGB-D... Following the default protocol, we partitioned the datasets into 4,845 images for training and 4,659 ones for testing. |
| Hardware Specification | Yes | Our model was implemented using the PyTorch framework and trained on NVIDIA GTX 2080ti GPU for a total of 100 epochs. Additionally, efficiency tests on a single RTX 3090 GPU at 192 × 320 resolution show our model achieves 101.7 GFlops, 41ms runtime, and 24.4 FPS. |
| Software Dependencies | No | Our model was implemented using the PyTorch framework. No specific version number for PyTorch or any other software dependencies is mentioned. |
| Experiment Setup | Yes | Our model was implemented using the PyTorch framework and trained on NVIDIA GTX 2080ti GPU for a total of 100 epochs. We adopted the SGD optimizer for training, with a momentum term of 0.95 and a weight decay term of 10-4. The initial learning rate was set to 1 × 10-3 and was halved during the plateau period. |